I'm doing some research on graph databases to see if they would fit my use case. Right now I'm looking at Janusgraph, but have also looked at ArangoDB and Neo4J a little.
I have a tree of nodes, but I was thinking of encoding version information in the edges, to create a sort of versioned database. For a given version, a traversal following the edges with that version specifier would output the correct version of that tree. This would produce a tree with n vertices but 100 (ish) edges per vertex (100n total).
Would there be any performance hit on traversal? Assume that there are indexes on vertexes (and edges, if that is relevant)